This study addresses the impact of projected changes to northeast monsoon on rice yield during rabi season (September–December) in Tamil Nadu by using a three‐step approach. First, coarse‐resolution global climate models that realistically capture the mean monsoon characteristics were selected. Second, lateral and boundary conditions taken from selected global models’ projections are employed to run a high‐resolution regional climate model. Third, climate variables from regional model being fed into panel data regression model. For different scenarios and for mid and end of century projections, in conjunction with projected rainfall, a comprehensive assessment is carried out to underscore the sensitivities of maximum and minimum temperatures under different stages of rice production, viz. vegetative, reproductive and maturity phases, and to the concept of growing degree days (GDD, cumulative heat effect). Irrespective of scenarios, in response to an increase in projected monsoon rainfall and surface temperature conditions, the regression model estimates an increase of rice yield of about 10–12% by mid‐century and 5–33% by the end of the century. In the regression model, the baseline coefficients were estimated from observed rainfall and temperature available from India Meteorological Department (IMD). The projected changes in rice yield, however, remain unchanged for baseline coefficients estimated from regional climate model outputs (forced by reanalysis products) rainfall and temperature. The robust results obtained here provide confidence to the findings.

The overall measure of farm‐level technical efficiency is generally used to derive recommendations for the use of individual inputs. In this paper, joint estimation is made of technical and individual input‐use (e.g. irrigation water productivity) efficiencies. This indicates that overall technical efficiency is not an indication of the efficiency level of all the individual inputs used. This is because the efficiency of individual inputs may vary and suggests that greater effort should be made to improve such efficiencies in comparison with overall technical efficiency. The model is applied to rice production in four tank‐irrigated districts in Tamil Nadu, India, which is one of the most important tank‐irrigated areas in India. The average technical efficiency is 62.8%, which indicates that in order to achieve the present level of production, 62.8% of the current level of input resources is sufficient. Average irrigation water productivity is estimated at ~34%, indicating that current output levels could be achieved with 66% less irrigation water. These findings also suggest the need for improvements in crop and water productivity. Thus, the paper makes a contribution in the form of a methodology development for possible adoption in future irrigation water productivity studies. Copyright © 2017 John Wiley & Sons, Ltd.